SAM-LST icon indicating copy to clipboard operation
SAM-LST copied to clipboard

Published 20 hours ago • verified publisher icon 11yxk

Results

Open XZNWU opened this issue 1 year ago • 2 comments

Thank you very much for sharing the code. We used the code you shared to train the model, but the prediction results we got were very poor, just like randomly generated results. Please tell me what causes this.

XZNWU avatar Oct 19 '23 01:10 XZNWU

I found that there seems to be a problem with the test code and the newly trained model cannot be added. Please tell me how to solve it.

XZNWU avatar Oct 19 '23 03:10 XZNWU

Hi , hope you will be fine.

Did you get the improved results ?

Also can you share your GPU memory space that was used during the experiment? I will be replicating same approach to fine tune SAM on custom medical dataset but my available memory space is only 8GB.

Thanks!

Raspberry-beans avatar Jan 10 '24 10:01 Raspberry-beans

Hi , I trained the model as the writer says , I also got the prediction results very poor , did you solve this problem? Thanks!

WenNa0616 avatar May 23 '24 06:05 WenNa0616

I got the Testing performance : mean_dice : 0.023857 mean_hd95 : 254.548986

WenNa0616 avatar May 23 '24 06:05 WenNa0616